* Kodama are forest spirits featured in Studio Ghibli's film: Princess Mononoke.

This is my project for the CIT Honors Research Program.  From my work on this project, I was awarded the "Excellence in Undergraduate Research" Award at Commencement.

I was part of the CMU Robomechanics Lab ( https://www.cmu.edu/me/robomechanicslab/ ) during the 2017-2018 Academic Year.  I worked with my professor, Aaron Johnson, and lab-mate, Praxis Bayes.

We were inspired by mountain goats and their great maneuverability on mountainous terrain.  We focused on studying the mechanics of their lower limb, specifically of their hooves.  We worked on designing mechanical hooves that conform to steep and uneven surfaces, and experimented with material, geometry, and fabrication methods

The Big Question:
How can we improve the mobility of legged robots on steep and rocky surfaces?

Research Goal:
To understand the principles underlying improved mobility on steep and rocky terrain.
Focus: Mechanics of the Feet.

The "Minitaur," a 4-legged robot developed by Ghost Robotics was used as a testing platform.

Hoof Design
The Hoof body was printed using the Form 2 3D Printer, and the sole was cast.

Ankle Design
Fabricated from bending spring steel rods.

This project was for 24-671: Electromechanical Systems Design with Professor Mark Bedillion

For our Senior Design Capstone Project, we set out to design a device that will enhance the quality of life of Parkinson's Disease (PD) Patients, specifically by addressing micrographia (difficulty of writing due to hand tremors).  In the ideation stage, we converged on the idea of a pen grip that will be able to sense and counteract tremors.  Within our team of 5, we split into subgroups to work on enclosure design and controls.

We designed and fabricated a series of prototypes:

• Tremor Generator: Because we did not have the opportunity to work directly with PD patients, we created our own generator that simulates tremors of different characteristic frequencies between 4-6.4 Hz.
• Tremor Sensor: We built a sensing system to test that we can pick up tremors accurately and at a sufficient sampling rate.  Later, it was integrated into the Anti-Tremor Generator.
• Anti-Tremor Generator: A device that can sense tremors and produce torques to stabilize the hand.

Unfortunately, the device did not work as we had planned. The issue was our incorrect assumption of the system dynamics. We presumed that the rotation in the wrist and hand can be mostly dampened by a single rotating motor. This idea was proven to be achievable in a computer simulation, as presented by the results of the Simulink Model. However, we ran into problems during system integration, where we were getting unwanted vibrations in other degrees of freedom. Although the motor behaved as wanted (as commanded by the IMP Controller), and moved in the general direction in an attempt to counteract the tremors, it unfortunately introduced torques in other axes.

Lastly, we would like to thank the course staff for their tremendous help and support!

Problem Statement:

• Environment

A multi-lane highway with cars moving in one direction at a constant (known) speed.

• Planner

Our car (non-holonomic) is trying to speed through the traffic!  Given a constant velocity that is greater than the traffic’s,  the start location and the goal location, the planner will create a plan (or return that there is no plan) for the car to navigate through the traffic from start to goal.

Algorithms & Techniques Involved:

• Explicit Graph, Plan then Execute
• Weighted A* Search
  • Heuristic:  Euclidean Distance
• Motion Primitives
  • Straight, Change Lane To Left, Change Lane To Right

Planner runs a Weighted A* Search on the States defined by (x, y, t).  Successors of each state are found by collision checking each motion primitive.  If no plan is found, the planner will return: “No Plan Found!”

Variables:

• Start Lane, Goal Lane
• Robot Speed
• Traffic Speed
• Percent Traffic
• Number of Lanes
• Road Length

Remaining Issues:

• More Realistic
  • Variable Speed
  • More Motion Primitives
• Safety
• Scalability & Flexibility
  • Be able to scale to different environments
• Precision
  • Increased precision in distance and time
• Move towards Real-Time Planning  

Task:
To design a bracket (astronaut's coat rack) that will support a suit clip loaded with 40 ± 1 pounds force directed downward, one time, without failing, for at least 10 seconds.  The bracket must not interfere with the forbidden zone (shown below in red), and the suit clip must not move more than 0.25 inches from its initial position.  The bracket will be mounted to the support structure on either side by socket cap screws.

Iteration, Iteration, Iteration!
Over 50 Prototypes were Laser Cut!      (click to see photos below)

A lot of work was put into achieving the final design.  Major modifications were made with mass minimization in mind.  Firstly, the bottom beam was shortened significantly to decrease both its magnitude of buckling and the part's mass.  Also, its width was increased at the center to resist buckling, tapering off on both sides.  On the other hand, the width of the top beam was decreased because it supports solely tensile axial loading.  The profile around pin supports were refined to reinforce specifically locations that experience critical contact stresses.

Estimated Mass of Final Design: 4.45 ± 0.05 grams 
* This is 0.025 % of the mass it is supporting!!!

Below is a slow-motion clip of my final design when loaded until failure.  The design failed where I had expected.  This prediction was also confirmed by FEA.

This is the CMU Aerospace Club's very first RC Plane!
We made it completely from scratch following a guideline found online.
We are currently working on it, hoping to finish before it snows!

We are only a few electronic components away from our first trial flight!

Because we are awesome CMU Engineers, of course we did not simply follow the instructions.  Some equations later, we altered the dimensions of the plane!  But this is all because we printed on the wrong page size... OOPS!

The most time-consuming process so far? .... Painting!
​We realized soon enough that spray paint was probably a great idea!